eugene@ames-lm.UUCP (Eugene Miya) (06/07/84)
I will be assuming the responsibility for electronic announcements
for the SF Bay area ACM SIGBIG (Special Interest Committee on
Large Scale Systems). We will broadcast to the local Bay area nets:
PARC, IBM, UCB, SU, LLNL, and the Usenet. All meetings are open.
I try provide short summaries to the net upon receipt of several
mail requests.
Last, evening we had Raul Mendez of the Naval Post Grad School
present his latest data comparing Cray XMP, Cyber 205, Fujitsu VP-100/200,
and Hitatchi H-810. They appear (CPU-wise) to be evenly matched.
Newer tests on I/O have not been analyzed yet. Raul described in some
detail the vectorizer on the VP. First, he wanted to clarify that
K. Muira is not a student of D. Kuck which was erroenously spread
when he got to this country [including this reporter on this net].
The vectorizer uses three different methods for vectorizing loops with IFs
dependent upon the qualitative nature of the loop: scatter-gather,
masking, etc. The meeting was held at Stanford U AEL.
The next meeting will be held at Pacific Sierra Research in Berkeley.
Speaker will be John Kileen, director of the National Magnetic Fusion
Energy CC at LLNL speaking on Supercomputers in Fusion Research.
Meeting will be July 11, Wed., 730pm at 2855 Telegraph. As the time
approaches, I will rebroadcast this.
Other news, we brought our 4-pipe Cyber 205 up this week at NASA.
It is not accessable (mail or file transfer) on this net, and reached
with difficultly for RJE on the ARPAnet. Anybody wishing time on the
205 should contact Technology Development of California in Santa
Clara as they run our 205.
--eugene miya
{hplabs,hao,research,dual,menlo70,icase,riacs}!ames-lm!eugene
NASA Ames Research Center